c语言超强删除注释的函数（linux、mac）

#include <stdio.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <sys/mman.h>
#include <unistd.h>
#include <sys/dir.h>
#include <map>
#include <string>
#include <fstream>
using namespace std;

#define  NULLCHAR  32       //ascii字符的32表示空字符

//删除注释.path:文件路径全名
void delComments(const char *path);

//判断是否为目录.path:文件路径全名
bool isDir(const char* path);

//删除文件夹下所有.c, cpp, .h的注释
void delComInDir(const char *path);

//删除文件夹下所有.c, cpp, .h的注释，这是递归方法
void delComInDir(const char *path, int recursive);

//取文件名的后缀名
const char *getSuffix(const char *path);

//取子串
char* substr(const char*str, unsigned long start, unsigned long end);


//删除注释类型
namespace CommentType
{
    //删除"特定函数".funStr:函数名称
    void delFun(const char *funStr, unsigned char *buffer, unsigned long size);

    //删除"/**/"型注释
    void delMuliCom(unsigned char *buffer, unsigned long size);
    
    //删除"//"型等但行注释
    void delSigleCom(const char *comStr, unsigned char *buffer, unsigned long size);
}




int main(int argc, const char * argv[])
{
    delComInDir("/Users/zctech/Desktop/test");
    //delComInDir("/Users/zctech/Downloads/temp/hjsg/Classes/network/websocket");
    printf("操作完成");
    return 0;
}




//删除注释.path:文件名
void delComments(const char *path)
{
    
    int fd = open(path, O_RDWR);
    
    //获取文件大小
    struct stat sb;
    fstat(fd, &sb);
    
    unsigned char *start = (unsigned char *)mmap(NULL, sb.st_size, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0);
    if(start == MAP_FAILED) /* 判断是否映射成功 */
    {
        printf("映射失败，文件过大或者没有权限");
        return;
    }
    CommentType::delSigleCom("//", start, sb.st_size);
    CommentType::delFun("CCLog", start, sb.st_size);
    CommentType::delFun("CCLOG", start, sb.st_size);
    CommentType::delSigleCom("#pragma", start, sb.st_size);
    CommentType::delFun("zctech::ZCLOGI", start, sb.st_size);
    CommentType::delFun("ZCLOGI", start, sb.st_size);
    close(fd);
    munmap(start, sb.st_size); /* 解除映射 */
    
}


//判断是否为目录.path:文件路径全名
bool isDir(const char* path)
{
    struct stat st;
    lstat(path, &st);
    return 0 != S_ISDIR(st.st_mode);
}



//删除文件夹下所有.c, cpp, .h的注释
void delComInDir(const char *path)
{
    unsigned long len;
    char temp[1024 * 1024];
    //去掉末尾的'/'
    len = strlen(path);
    strcpy(temp, path);
    if(temp[len - 1] == '/') temp[len -1] = '\0';
    
    if(isDir(temp))
    {
        //处理目录
        int recursive = 1;
        delComInDir(temp, recursive);
    }
    else   //输出文件
    {
        //找出.h和.cpp文件
        if(0 == strcmp("h", getSuffix(temp)) || 0 == strcmp("cpp", getSuffix(temp)) || 0 == strcmp("c", getSuffix(temp)))
        {
            printf("%s\n", temp);
            
            delComments(temp);
            
        }
        
    }
    
}


//遍历文件夹de递归函数
void delComInDir(const char *path, int recursive)
{
    DIR *pdir;
    struct dirent *pdirent;
    char temp[1024 * 1024];
    pdir = opendir(path);
    if(pdir)
    {
        while((pdirent = readdir(pdir)))
        {
            //跳过"."和".."
            if(strcmp(pdirent->d_name, ".") == 0
               || strcmp(pdirent->d_name, "..") == 0)
                continue;
            sprintf(temp, "%s/%s", path, pdirent->d_name);
            
            //当temp为目录并且recursive为1的时候递归处理子目录
            if(isDir(temp) && recursive)
            {
                delComInDir(temp, recursive);
            }
            else
            {
                //找出.h和.cpp文件
                if(0 == strcmp("h", getSuffix(temp)) || 0 == strcmp("cpp", getSuffix(temp)) || 0 == strcmp("c", getSuffix(temp)))
                {
                    printf("%s\n", temp);
                    delComments(temp);
                }
            }
        }
    }
    else
    {
        printf("opendir error:%s\n", path);
    }
    closedir(pdir);
}




//求子串
char* substr(const char*str, unsigned long start, unsigned long end)
{
    unsigned long n = end - start;
    static char stbuf[1024 * 1024];
    strncpy(stbuf, str + start, n);
    stbuf[n] = 0;                           //字串最后加上0
    return stbuf;
}


//取文件名的后缀名
const char *getSuffix(const char *path)
{
    char *ptr;
    long pos;
    ptr = strrchr(path, '.');    //最后一个出现c的位置
    if(NULL == ptr)
    {
        return "";
    }
    pos = ptr - path;             //用指针相减 求得索引
    
    return substr(path, pos + 1, strlen(path));
}



//删除"//"型注释
void CommentType::delSigleCom(const char *comStr, unsigned char *buffer, unsigned long size)
{
    //如果要删除但行注释先要删除多行注释/**/
    CommentType::delMuliCom(buffer, size);

    unsigned long i = 0;
    bool isCom = false;
    bool isStr = false;
    while (i < size)
    {
        if(!isStr && 0 == strcmp(comStr, substr((char *)buffer, i, i + strlen(comStr))))
        {
            isCom = true;
        }
        else if(!isStr && isCom && '\n' == *(buffer + i))
        {
            isCom = false;
        }
        else if(!isCom && !isStr && '\"' == *(buffer + i) && '\\' != *(buffer + i - 1))
        {
            isStr = true;
        }
        else if(!isCom && isStr && '\"' == *(buffer + i) && '\\' != *(buffer + i - 1))
        {
            isStr = false;
        }
        
        if('\n' == *(buffer + i))
        {isStr = false;}
        
        //移除注释
        if (isCom && !isStr)
        {
            *(buffer + i) = 32;
        }
        
        i++;
    }
}

//删除"/**/"型注释
void CommentType::delMuliCom(unsigned char *buffer, unsigned long size)
{
    //判断是否为 【/**/】 型注释, 注意当是字符串的时候不替换
    unsigned long i = 0;
    bool isCom = false;
    bool isStr = false;
    
    while (i < size)
    {
        if(!isStr && '/' == *(buffer + i) && '*' == *(buffer + i + 1))
        {
            isCom = true;
        }
        else if(!isStr && isCom && '*' == *(buffer + i) && '/' == *(buffer + i + 1))
        {
            isCom = false;
            *(buffer + i) = *(buffer + i + 1) = 32;
        }
        else if(!isCom && !isStr && '\"' == *(buffer + i) && '\\' != *(buffer + i - 1))
        {
            isStr = true;
        }
        else if(!isCom && isStr && '\"' == *(buffer + i) && '\\' != *(buffer + i - 1))
        {
            isStr = false;
        }
        
        if('\n' == *(buffer + i))
        {isStr = false;}
        
        //移除注释
        if (isCom && !isStr)
        {
            *(buffer + i) = 32;
        }
        i++;
    }
}

//删除"特定函数".funStr:函数名称
void CommentType::delFun(const char *funStr, unsigned char *buffer, unsigned long size)
{
    
    unsigned long i = 0;
    bool isStr = false, isFun = false;
    int left = 0, right = 0;    //左右括号
    
    while (i < size)
    {
        if(!isStr && 0 == strcmp(funStr, substr((char *)buffer, i, i + strlen(funStr)))
           
           )
        {
            isFun = true;
        }
        else if(!isFun && !isStr && '\"' == *(buffer + i))
        {
            isStr = true;
        }
        else if(!isFun && isStr && '\"' == *(buffer + i))
        {
            isStr = false;
        }
        
        
        if(isFun && '(' == *(buffer + i))
        {
            left++;
        }
        else if(isFun && ')' == *(buffer + i))
        {
            right++;
            if(left == right)
            {
                *(buffer + i) = NULLCHAR;
                isFun = false;
                unsigned long j = i + 1;
                while (' ' == *(buffer + j) || ';' == *(buffer + j))
                {
                    *(buffer + j) = NULLCHAR;
                    j++;
                }
            }
        }
        
        //移除注释
        if (isFun && !isStr)
        {
            *(buffer + i) = NULLCHAR;
        }
        
        i++;
    }
}